Pakistan launches AI-powered climate intelligence platform at COP31


  • Zaki Abbas
  • November 17, 2026
Residents use a boat while others wade through the rising floodwater, following rains and floods during the monsoon season on the outskirts of Bhan Syedabad, Sindh, September 8, 2022. — REUTERS

ISLAMABAD: As Pakistan faces frequent climate disasters due to global warming, a local climate tech firm has launched an AI-powered climate intelligence platform, Nigraan, to turn data into decisions in a bid to empower communities and decision-makers amid a debate on dangers AI infrastructure poses to environment.

Launched at COP31 in Antalya by AI Geo Navigators, Nigraan — an acronym for the National Integrated Geo Risk Alert and Atmospheric Network — was trained on “28 million rows over 18 months”, with the sole intention of filling information gaps.

Its creators say the model integrates weather data, satellite observations, and AI-based hazard assessments to deliver hyperlocal weather insights and climate risk intelligence, from national and regional levels down to individual locations and sites.

In an interview with Hum News Digital, AI Geo Navigators CEO Asim Javid said the AI-powered platform will provide real-time weather data with a focus on localised information. The model uses remote sensing, GIS, artificial intelligence and crowdsourcing to gather data, said Javid, adding that Nigraan will cover the entire country, from the Himalayas to the Arabian Sea, and provide climate solutions with “precise information for precision locations”.

According to Javid, the model was launched at the Pakistan Pavilion at COP31 because they wanted to have a global audience, particularly in the Global South countries that could not deploy such technologies due to resource constraints.

The launch came after Turkiye launched the “Antalya Artificial Intelligence Pledge” at COP31 to make artificial intelligence part of the climate solution in a sustainable manner. The pledge, which provides a voluntary framework, emphasises the use of low-emission energy for AI data centres, transparency regarding its environmental impact, and financial support for AI solutions across the developing world.

But the concerns about the harmful impacts of AI persist even as the world increasingly turns to AI for climate information and solutions. “The question isn’t simply whether AI is good or bad for the climate. It’s whether people can rely on what they’re told by AI and about it,” said Charlotte Scaddan, Information Integrity Senior Advisor with the UN Department of Global Communications at the launch of Priorities for Action on AI and Climate Information Integrity in Antalya.

Javid acknowledges harmful impacts of AI, but believes the benefits outweigh the risks. He said AI has made it easier for them to read massive datasets with little effort and that swift action without AI was “impossible”. For instance, statistical models for flood predictions would require a couple of days to process data, but AI can do it in two hours, he said, adding that global climate models were also using AI.

On a smaller level, AI can help communities to adopt a sustainable lifestyle, he said, referring to a rain harvesting application RainClove by AI Geo Navigators. The AI-powered application helps people harvest rainwater in Islamabad by calculating “annual harvest in litres, optimal tank size, local rainfall trends and flood insights”.

Is AI helpful for climate action?

Dr Zubair Khalid, who teaches at Lahore University of Management Sciences (Lums), said AI can help in data integration but emphasised the importance of indigenous models for accurate information.

Benefits outweigh the risks if systems are validated against local data and their uncertainty is communicated honestly, he said, adding that a forecast that was confident but wrong could do more harm than no forecast.

“AI can help most where we lack dense ground measurements, which is the case in most Pakistani cities. It can combine satellite data, weather models and sparse sensor readings to forecast floods, heatwaves and smog episodes at a much finer scale than before,” he said.

“The main risks are models trained on data from elsewhere, weak ground truth for validation, and warnings that do not reach the most vulnerable people in a language and format they can act on,” said Dr Khalid, who was part of the team that was awarded the Gordon Bell Prize for building a “smart climate model”.

In response to a question about energy-intensive AI infrastructure, the Lums professor said most climate adaptation use cases did not need the largest models. “Smaller, task-specific models, often running on edge devices, can deliver most of the benefit at a fraction of the energy cost,” he said, calling for transparency in the energy use of AI systems.

As AI finds a friend in COP31, experts warn that the climate talks need to address this crucial issue as unchecked expansion of energy-guzzling data centres poses more risks than benefits to society. The International Energy Agency quoted by Reuters has estimated that global data centres would consume about 945 terawatt-hours of electricity annually by 2030, producing the equivalent of more than 300 million metric tons of carbon dioxide.

Sara Shaw, 350.org director for campaigns and networks, told Hum News Digital that the COP process needed to start putting AI on the agenda. “But as it stands, the AI pledge risks turning into a dangerous smokescreen that gives Big Tech the license to pollute while still devouring land, water and energy,” she said, adding that the pledge fails to commit that no new fossil fuel infrastructure would be built for AI, while weakly referencing “zero and low-emission energy” instead.

You May Also Like

WhatsApp